Generally, in case that an audio signal, and more particularly, the audio signal has strong characteristics of a speech signal, linear predictive coding (LPC) is performed on the audio signal. A linear predictive coefficient generated by linear predictive coding is transmitted to a decoder. Subsequently, the decoder reconstructs the audio signal by performing linear predictive synthesis on the corresponding coefficient.